πέντε, ἕξ, ἑπτά, and their compounds
(Phryn. Ecl. 386, Phryn. Ecl. 387, Phryn. Ecl. 388, Moer. π 63, Antiatt. ε 122)

D. General commentary

Several entries in Atticist lexica – probably drawing on a discussion that ultimately goes back to Aelius Dionysius (see the apparatus at B.3 and B.5) – address compounds whose first constituent is derived from the numerals πέντε ‘five’ (A.1, A.2, A.4) and ἑπτά ‘seven’ (A.5); an entry in the Eclogue (A.3) further discusses the compounds of ἕξ ‘six’. The focus of these entries may be summarised as centring on two main issues.

The first concerns whether the final vowel of the numeral in the first constituent should be elidedElision (πεντ-, ἑξ-, ἑπτ-) or retained (πεντα-, ἑξα-, ἑπτα-) when it precedes a second constituent beginning with a vowel – usually -ετ- from ἔτος (< ϝέτος) ‘year’. This is the focus of the first entry of Eclogue (A.1), which contrasts the unelided forms πενταετηρικός ‘quinquennial’ and πενταετηρίς ‘festival celebrated every fifth year’ with the elided variants πεντετηρικός and πεντετηρίς; of Eclogue 388 (A.3), which discusses ἑξάπηχυ ‘six cubits long’ and ἑξαέτης ‘of six years’ (but also ἑξάπλεθρον ‘six plethra long’), in contrast with their elided counterparts; and of the Antiatticist (A.5), which deals with the elided adjective ἑπτέτης ‘of seven years’ and implicitly opposes it to the unelided form ἑπταέτης. (Note that a similar discussion also involved compounds of δέκα, which within Atticist lexicography, however, are addressed only in Moer. δ 13Moer. δ 13: δεκέτης Ἀττικοί· δεκαέτης Ἕλληνες, ‘Users of Attic [employ] δεκέτης; users of Greek [employ] δεκαέτης’; cf. also Lobeck 1820, 406–11).

The second issue emerging from these entries concerns the shape that πέντε should take in compounds, a topic also addressed by other ancient scholars as part of a broader debate on the correct forms of this and other numerals in compounds and derivatives (see B.4, B.5, and partly B.6). Phrynichus (A.2) identifies πέντε- as the correct compositional form, as opposed to πέντα-. This is also the focus of Moeris’ entry (A.4), which additionally discusses the elided form πεντετηρίς. While the first of these two issues might appear to concern merely Kompositionsfuge (i.e. the transition from the first to the second constituent), it in fact also involves the formation of Greek numerals in relation to the parent language and their dialectal evolution. The following commentary therefore provides an overview of the history of πέντε, ἕξ, and ἑπτά before turning to the lexemes discussed by Atticist lexicographers and to the distribution of the various by-forms across different dialects and historical phases of Greek.

The Greek numeral πέντε (πέμπε in Aeolic) regularly results from Indo-European *penk^we (see e.g. Sihler 1995, 413). πέντε also occurs in the numerals ‘fifteen’ (πεντεκαίδεκα), ‘twenty-five’ (εἰκοσπέντε), and similar forms in the following decades. By contrast, πεντήκοντα ‘fifty’ shows a lengthened vowel, which has been explained in different ways: by analogy with τετράκοντα/τετρώκοντα ‘forty’, where the long vowel is original (Szemerényi 1960, 129–30); as compensatory lengthening from an original *penk^we-dḱomth₂ (Sihler 1995, 418); or as the reflex of a laryngeal, itself resulting from a glottal stop (*penk^we-h₁ḱomth₂: Kortlandt 1983, 98; Blažek 2000, 102; Rau 2009, 20). No unequivocal forms of ‘five’ are so far attested in Mycenaean (though see Waanders 1992, 385 for some speculation concerning personal names that may be compounded with this numeral). The Greek numeral ‘six’ is ἕξ, also dialectically attested as ϝέξ, which points to PIE *s(u̯)ek̑s (see DGE s.v.; Rothstein-Dowden 2022, 139). The etymology of the underlying PIE form, however, is much debated (overviews in Sihler 1995, 413; Rothstein-Dowden 2022, especially with regard to the outcomes in Indo-Aryan). ἑπτά is the regular outcome of Indo-European *septḿ̥.

When used as the first constituent of compounds, πέντε usually takes the form πεντα-, where -α- is analogicalAnalogy on the compositional forms of the flanking numerals ‘four’ (τετρα-, whose /a/ is the regular outcome of a syllabic resonant: < *k^wetu̯r̥-, see Sihler 1995, 412) and ‘six’ (ἑξα-, itself analogically modelled on ἑπτα-: see below). The first constituent πεντε-, though formally identical to the original numeral, is more likely a later analogical remodelling of the more common πεντα- rather than a genuine relic. This is suggested by the relatively late appearance of compounds in πεντε-, which are unattested before 5th-century Attic texts (see below). A further compositional form is πεμπ-, always found before /a/ and /o/ (e.g. πεμπώβολον ‘five-pronged fork’, Hom.+). This form, which shows the expected labial outcome of the original labiovelar *k^w before /a/ and /o/, was gradually replaced by πεντ- (see πεντώβολος ‘worth five obols’, Ar. Eq. 798; cf. Waanders 1992, 372). The Greek grammatical tradition also addressed the origin of πεντα- and πεντε-, debating whether the α in these and similar numeral compounds derives from the corresponding adverbs (e.g. πεντάκις ‘five times’) or from the numerals themselves. The doctrine reported by the Etymologicum Gudianum (B.4) and attributed to Herodian (see the apparatus ad loc.) correctly identifies the source of πεντα- in the numeral (πέντε) rather than in the adverb (πεντάκις): the fact that the Attic compounds take πέντε- as their first constituent is adduced as a evidence in support of this explanation. The opposite explanation is instead defended in the etymological tradition reported in the Etymologicum Symeonis, which has parallels in the Genuinum and the Magnum (see B.6). As regards ‘six’, when used in compounds ἕξ may retain its original form (ἑξ-, the form prescribed by Phrynichus in A.3; also attested as ἑκ- and ἑσ-), but the normal form is ἑξα-, analogical on ἑπτα-. The latter is the only compositional form of ἑπτά. A tradition reported in the etymologica (B.6) endorses the forms in ἑξα- while proscribing those in ἑξ- on the grounds that no compositional first constituent preserves a double consonant.

Returning to the Atticist sources, a preliminary caveat is in order. Any analysis of the distribution of forms in ἑπτα-/ἑπτε-, ἑξα-/ἑξ-, and πεντα-/πεντε- is inherently compromised by the fact that many of their occurrences show variation in the manuscript tradition (on the issue of textual variations, see already Lobeck 1820, 406; 411). In what follows, this will be illustrated with references exclusively to the attestations of πενταετηρικός/πεντετηρικός and πενταετηρίς/πεντετηρίς, but the same considerations apply to all the other lexemes in the remainder of the commentary. Statements such as ‘this form is first attested in author X’ should therefore be understood as referring to the state of the evidence as it can be reconstructed through the TLG, and as resting on the reasonable assumption that modern editors have exercised their best judgment in establishing critical texts. The original state of affairs is, unfortunately, difficult – if not impossible – to reconstruct.

It is convenient to begin with compounds whose first constituent is the numeral ἑπτα-, since their treatment in ancient sources is more succinct. The focus of the entry in the Antiatticist (A.5), as already noted, is on the admissibility of the elided form ἑπτέτης, implicitly contrasted with the unelided form ἑπταέτης (for the alternative accentuation ἑπταετής, and for the issue of the recessive stress in compounds of this kind in relation to other compounds in -ής, see Meissner 2006, 199–200, 205–6; Probert 2006, 88 with n. 14 discusses these forms in the wider context of Bartoli’s and Vendryes’ laws; Dieu 2022, 409–11 considers the recessive accentuation typical of Attic and provides an ample bibliography on the matter). ἑπταέτης is attested from the Odyssey onwards (C.1 and two further loci), where the hiatus is likely justified by the presence of a still operational /w/ in the second constituent (ϝέτος). In Attic, the -α- of ἑπτα- is regularly elided before another vowel. The form ἑπτέτης is used by Chionides (C.4), though only in a contextless one-word quotation cited by the Antiatticist (A.5), and by Aristophanes (C.6, where an original ἑπταέτης would also have been metrically possible: see Rutherford 1881, 489; cf. also the feminine ἑπτέτις in Ar. Th. 480). If we can trust the texts as edited by modern editors, the elided form ἑπτέτης remains an Attic peculiarity, further attested in Plato (Grg. 471c.2), but never in inscriptions or papyri. ἑπταέτης, by contrast, is the normal variant not only in non-Attic authors (see e.g. Hp. Carn. 13.4 Joly = 8.602.5–6 Littré; Hellanic. FGrHist 4 F 168b.2), but also in Aristotle (GA 747b.25; HA 577b.20) and across the various registers of the written koine. Particularly telling, in this respect, is its occurrence in the Septuagint (Id. 6.25), followed by Diodorus Siculus (11.38.7), Philo of Alexandria (De sobrietate 8.8: ἑπταέτιν), Plutarch (e.g. Lyc. 16.4.7), and many others. In conclusion, the preponderance of ἑπταέτης in texts of different periods and registers, including inscriptions, suggests that this form was maintained as the ‘normal’ one in contrast to the elided ἑπτέτης. We may therefore speculate that the latter (like the elided first constituents from πέντε – see below) was praised by the Atticists by virtue of both its Attic pedigree and its rarity. In this scenario, the precise aim of the Antiatticist (A.5) is unclear. If its purpose was the promotion of ἑπτέτης, a far more authoritative model than Chionides (C.4) would have been Aristophanes (C.6, on which see also Su. ε 2872). The issue, however, is likely to remain unsolved (Bagordo 2014, 48 cautiously considers the possibility that the original form in Chionides may have been the feminine ἑπτέτις, for which, however, there is another authoritative attestation in Ar. Th. 480).

Turning to compounds of πέντε, the first two variants addressed by Phrynichus in Ecl. 386 (A.1) – πενταετηρικός and πεντετηρικός (both ‘of five years’) – are attested only in post-classical texts. πενταετηρικός first occurs in the Septuagint (C.13, no variants recorded in Hanhart–Kappler), and later in a number of koine texts, such as D.S. 3.43.1 (albeit with the variant πεντετηρ- in cod. L), Str. 16.4.18 (no variants recorded by Radt), Ios. AJ 15.268.1 (no variants), and several others. πεντετηρικός first occurs in Strabo (C.14, where Radt reports no variants) and is later attested only in Josephus (AJ 16, pref. 26, where the manuscripts also report the variant πεντετηρ-), Cassius Dio (51.1.2, where Freyburger–Roddaz report no variants), and Phot. π 123 (Παναθήναια· ἀγὼν πεντετηρικὸς Ἀθήνησιν, ‘Panathenaea: A quinquennial festival in Athens’). The variant with /a/ is therefore the more common in Greek down to the Byzantine period (it is also the title of an oration by Themistius, Oration 8 Downey–Schenkl), whereas the Attic form is a rarity. Attestations of both forms in inscriptionsInscriptions are likewise post-classical (πενταετηρικός, the more frequent form, is mostly attested in the Roman period, though see e.g. I.Kaunos 62.6 [after 167 BCE] for a Hellenistic attestation, as well as the earlier occurrence of πενταετηρίς e.g. in F.Delphes 3.3.215[3].29 [245 BCE]; πεντετηρικός, by contrast, is very rare, with its earliest attestation possibly in I.Milet 3.1366.13–4, though the form is almost entirely restored; for πεντετηρίς, see e.g. ID 104(10).13 [355/4 BCE]).

The existence of the form πεντετηρικός in Attic is suggested by the frequency of the feminine noun (and adjective) πεντετηρίς (‘[festival] celebrated every fifth year’), also addressed by Phrynichus in A.1. Besides Thucydides (C.7, where, however, some MSS also have πενταετηρ-), the form is found in D. 24.125.6 (see further the attestations in Hdt. 3.97.4, where cod. D has πενταετηρ-; 4.94.2, and 6.87.1; the occurrence in Lycurg. In Leocratem 102.3 Conomis is a modern correction for the manuscripts’ πενταετηρ-). In Attic inscriptions, πεντετηρίς is attested from the 5th century BCE onwards (e.g. IG 1³.1468.1 [Athens, 425/5 BCE]). The comic compound πρωκτοπεντετηρίς (‘quinquennial debauchery’) of Ar. Pax 876 is especially informative, since it is metrically guaranteed and thus confirms the existence of the variant with the elision of α in the Kompositionsfuge. πεντετηρίς is also amply attested in later texts, including Atticising authors, e.g. Aristid. 1.230 Lenz–Behr (= 13.159.11 Dindorf); Poll. 8.107; Ath. 5.198b; Lib. Or. 46.22.3. The textual tradition, however, is often perturbed. For instance, at Aristid. 1.230 Lenz–Behr (= 13.159.11 Dindorf) the reading πεντετηρίδα is reported by various manuscripts (see the examples in Dindorf’s apparatus) as well as by Thomas Magister’s quotation of the passage (B.7), whereas one manuscript, cod. Ambrosianus B 69 sup., has the variant πενταετηρίδα (which, on the basis of this manuscript, was printed in the Aldine princeps). Similarly, Pollux’s πεντετηρίδας at 8.107Poll. 8.107 is transmitted as πενταετηρίδας in the manuscripts of Bethe’s family II (= ‘family b’ in the terminology adopted by Cavarzeran 2025, 59–60; see also entry Iulius Pollux, Ὀνομαστικόν (Onomasticon), section B). Turning to πενταετηρίς, the form first occurs in Pindar (C.3 and O. 10.57, N. 11.27) and is pervasive in post-classical prose (e.g. D.S. 5.32.6, where, however, cod. C has the variant πεντετηρ-; Str. 11.4.3; Ios. AJ 15.269.2). As anticipated, the suspicion that transmission and modern editorial practices may not reflect the original picture of the distribution of the two variants is strong: it suffices here to note the occurrence of the form with /e/ at Arist. Ath. 54.7.3 and 54.7.4 (no variant reported in all the major editions), and of that with /a/ at Pol. 1308b.1 (again with no variants reported). At any rate, the Attic form πεντετηρίς gradually died out in Greek: it is attested only twice in papyriPapyri (the earliest instance occurring in P.Ryl. 4.562.10 (= TM 2418) [Philadelphia, 251 BCE]). This may explain why Hesychius (B.1) includes a lemma πεντετηρίς in his lexicon and glosses it with the – by that time more common – by-form πενταετηρίς. It may be noted in passing that the cognate form πενταέτηρος ‘of five years’ occurs several times in Homer (Il. 2.403 and passim), though it is rare in later texts. This, again, explains its inclusion in Hesychius’ lexicon (B.2; no variant πεντέτηρος is attested in Greek texts).

The immediately subsequent entry in the Eclogue (A.2) addresses the question of whether the first constituent of compounds of ‘five’ should be πεντα- or πεντε- by focusing on two further adjectives, πεντάμηνος ‘five months old; of five months’ and πεντάπηχυς ‘five cubits long’, here lemmatised in the neuter (πεντάμηνον, πεντάπηχυ). The picture presented by modern editions is one in which the variants with /a/ are far more frequent throughout the diachrony of Greek than those with /e/ (the latter are never attested in inscriptions). πεντάμηνος first occurs in Hippocrates’ Epidemics (C.10, no variants), but it is also the sole reading at Men. Epit. 1116 (where van Herwerden proposed emending it to πεντέμην-, a correction now accepted by Kassel and Schröder in PCG vol. 6,1), and it remains in fact the only form in use in post-classical texts (including Plutarch, Galen, Ptolemaeus, and a wide range of astronomical works) down to the Byzantine age. Conversely, attestations of the form πεντέμηνος in literary texts are confined to a comic adespoton (C.8), transmitted by a 2nd- or 3rd-century CE papyrus, P.Oxy. 2806 (= TM 59788), a fragment previously tentatively ascribed to Cratinus by Austin, CGFP fr. 76 (for the view that it is better regarded as an adespoton, perhaps but not certainly by Cratinus, see most recently Henry 2013). The papyrus shows no variants or corrections, and the form with /e/ is therefore textually secure. Outside this text, πεντέμηνος is guaranteed only in the lexicographical and erudite works that discuss it, since it is explicitly contrasted with πεντάμηνος (besides A.2 and A.4, see B.3, B.4, B.5, B.6, and B.7). πεντέμηνος is not attested in inscriptions and papyri, where πεντάμηνος is the standard form.

The situation is similar for πεντάπηχυς/πεντέπηχυς. On the basis of modern editions, the variant with /a/ is the only form in use from Herodotus (9.83.9) onwards, extending to 4th-century BCE Attic (Thphr. HP 2.6.95, 4.9.1, 9.4.2), koine texts (e.g. LXX 1Pa. 11.23.2; D.S. 17.95.2; Ios. AJ 3.109.2, etc.), and Atticising authors as well (Poll. 4.170Poll. 4.170; Philostr. VA 2.4.3, 4.16.12; Her. 31.3.2). Conversely, πεντέπηχυς surfaces only in lexica (besides A.2, see B.3, B.5, and B.6), a finding confirmed by its absence from inscriptions and papyri (where πεντάπηχυς is the sole variant). To the forms already discussed by Phrynichus – πεντετηρίς and πεντέμηνον – Moeris (A.4) further adds πεντέκλινον ‘with five couches’, a neuter or masculine adjective in the accusative. Once again, the variant with /e/ is attested exclusively in lexicographical sources (see B.3, B.4, B.5, B.6, and B.7). πεντάκλινος, by contrast, occurs in Aristotle (Mir. 842b.22), as well as in a number of both koine (e.g. Phylarchus FGrHist 81 F 44.29) and Atticising authors (e.g. Poll. 1.79Poll. 1.79, 6.7Poll. 6.7; Ath. 4.142c, etc.). Neither form is attested in inscriptions; however, πεντάκλινος does occur in Hellenistic papyri (e.g. P.Cair.Zen. 1.59054.34–5 (= TM 2296) [Alexandria, 257 BCE]).

In conclusion, the Atticist lexicographers promoted the compounds in πεντε- because some of them were attested in classical Attic texts, but also – certainly – because of the rarity (and hence markedness) of all these forms in relation to those in πεντα-. The latter were the norm in non-Attic dialects as well as in Post-classical Greek, and they begin to surface already in Attic New Comedy. This picture is confirmed by the behaviour of other compounds of πέντε, on which see Lobeck (1820, 413).

As concerns compounds of ἑξ ‘six’, the only Atticist lexicon to address them is the Eclogue (A.3). Phrynichus condemns forms in ἑξα-, which he also judges to be typical of medical languageMedical language, and recommends ἑξ- as the only correct compositional form. The first proscribed form, ἑξάπηχυς, already occurs in Herodotus (C.5), while the recommended form in ἕξ- is attested in a one-word quotation from Sophocles (C.9), reported in the etymologica (B.6, which further attributes ἕξπους ‘six feet long’ to Plato Comicus, fr. 242). The attestation of ἕξπηχυς in Xenophon (C.12), retrievable via the TLG, is a modern correction by Cobet (accepted by Marchant in the Oxford edition and by Masqueray in the Belles Lettres edition) for the unanimously transmitted ἑξάπηχυ (retained by Hude in his edition). Apart from the Sophocles fragment and this contested passage of the Anabasis, forms in ἑξα- are the only ones retrievable through a TLG search at all stages of the language. The same picture emerges from inscriptions, where ἕξπηχυς is never attested and ἑξάπηχυς is the sole form (though attested only from the Hellenistic period onwards: see e.g. ID 290.216 [246 BCE]; the same applies to papyri). The occurrence of ἑξάπηχυς in Hippocrates (C.11) may underlie Phrynichus’ attribution of the ἑξα- forms to the language of doctors. The elided form of ἑξαέτης ‘of six years’, ἑξέτης, already occurs in the Iliad (C.2; the same collocation, ἑξέτε’ ἀδμήτην, also in Il. 23.655) and continues in Pi. N. 3.49, Ar. Nu. 862, Pl. Lg. 793e.4, etc. The extended form ἑξαέτης, perhaps exploited for metrical convenience in the Odyssey (C.1), gains ground in the Hellenistic period (see Theocr. 14.33) and remains the standard form in Post-classical Greek across all registers (see e.g. D.H. 3.57.1, Ios. AJ 19.355.1). This shows that Phrynichus’ precepts had no impact on contemporary literary practice and may also explain why the etymologica (B.6), while generally considering the Attic forms acceptable, nevertheless proscribe the compounds in ἑξ-. Inscriptions, too, attest only the form ἑξαέτης (with no occurrences in Attica, however), and the same applies to papyri. Finally, ἑξάπλεθρος (‘six plethra long’) is attested only with /a/, beginning with Hdt. 2.149.3; it is an exceedingly rare form (see further Arr. 10.7.2) and is not found in inscriptions or papyri. The forms without /a/, which Phrynichus recommends, are not attested anywhere.

E. Byzantine and Modern Greek commentary

In Byzantine and Medieval Greek, the numeral ‘five’ remains πέντε. In Medieval Greek, ‘six’ becomes ἕξι or ἕξε (the latter perhaps analogical to πέντε: cf. CGMEMG vol. 2, 1247), while the classical form ἕξ is retained mostly in high-register or formal texts. As for ‘seven’, the ancient form ἑπτά is preserved in Byzantine literary language, but the phonological variant ἑφτά, resulting from manner dissimilation in the occlusive /p/, is attested in literary and documentary texts in Medieval Greek (see CGMEMG vol. 2, 1247). This variant never appears in compounds, where the standard form remains ἑπτα-. The dominance of the non-Attic variants with /a/ in Ancient Greek becomes complete in Byzantine Greek, where the forms in πεντε- appear – when they appear at all – almost exclusively in lexica, grammars, and scholia. No form in πεντε- survives in Medieval and Modern Greek.

F. Commentary on individual texts and occurrences

N/A